Davidmanheim comments on Non-Adversarial Goodhart and AI Risks